一种基于BC-RBFNN的高填路堤地基沉降预测模型

A Foundation Settlement Prediction Model Based on BC-RBFNN for High-Filled Embankment

针对高填路堤地基沉降难以预测这一技术难题,对其影响地基沉降的主要因素进行了分析,根据各因素之间存在的高度非线性,结合BC-RBFNN(基于聚类分析径向基函数神经网络)非线性拟合的特点,提出一种基于BC-RBNN模型对高填方地基沉降进行预测.运用施工期路基沉降实测资料,对神经网络模型进行学习、训练和仿真,得出仿真值与实测值非常相似,从而得出基于BC-RBFNN模型在高填路堤地基沉降预测中具有很好的实用效率.

Aimed at the technical difficulty that foundation settlement of high-filled embankment is hard to predict,the author analyzed the main factors influencing foundation settlement,and proposed a BC-RBFNN（Based on Clustering Radial Basis Function Neural Network） model to predict the foundation settlement for high-filled embankment according to the non-linear of each main influencing factor.Then,the author used the model to learn,train and emulate based on the monitoring data of settlement during construction.The results indicate the emulating data is extremely similar to the monitoring data and validate the useful efficiency of BC-RBFNN model applying on the foundation settlement prediction of high-filled embankment.